Ann. Rev. Cell BioI.

Copyright

1985 by Annual

1985. 1: 289-315
Reviews Inc.

ANNUAL REVIEWS

Further

All rights reserved

Quick links to online content

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

Andrew W. Murray and Jack W. Szostak Department of Molecular Biology, Massachusetts General Hospital, Boston, Massachusetts 021 14

CONTENTS
INTRODUCTION ......................................... A MODEL FOR CHROMOSOME BEHAVIOR DURING MITOSIS AND MEIOSIS ......................................

289 291 291 291 291 293 294 295 295 295 297 297 298 299 300 301 302 306

The

Interphase ......................................................................................................................................... Metaphase ........................................................................................................................................

Mitotic Cell Cycle

.................................................................................................................

Meiosis

Anaphase ...........................................................................................................................................
................................................................................................................................................

EXPERIMENTAL EVIDENCE .....................................................................................................................

The Mitotic Cell Cycle .................................................................................................................

Time of kinetochore duplication ..................................................................................................... . Centromeric DNA replication ........................................................................................................ . DNA replication and catenation ..................................................................................................... Prometaphase chromosome movement ............................................................................................ Micromanipulation of metaphase chromosomes ............................................................................ . Anaphase ........................................................................................................................................... Artificial chromosomes .....................................................................................................................

Mii:....:..::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Pairing and disjunction .................................................................................................................... Yeast meiotic mutants .................................................................................................................... ..

Drosophila meiotic mutants ............................................................................................................ Meiosis I versus meiosis II .............................................................................................................

306 308
309 311 311

EVOLUTION OF MEIOSIS ..........................................................................................................................

INTRODUCTION The faithful inheritance of genetic information depends on the orderly segregation of chromosomes in mitosis and meiosis. Mitotic segregation produces genetically identical daughter cells, while meiotic segregation produces cells that contain only one member of each chromosome pair that 289 0743-4634/85/1 1 15--0341$02.00

290

MURRAY & SZOSTAK

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

was present in the parental cell. Chromosome segregation is an extremely accurate process: Chromosome loss in mitosis in yeast occurs once in every 105 divisions (1), while aberrations in meiotic chromosome segregation occur once in every 104 meioses in both yeast (2) and Drosophila (reviewed in 3). Recent reviews have discussed mitosis with reference to mechanisms of chromosome movement (4-6), the role of membranes in the behavior of the mitotic spindle (7), and the structure and behavior of kinetochores (8). The genetic control of meiosis has been reviewed for yeast (9), Drosophila (3), and other organisms (10). Ultrastructural aspects of meiosis have also been reviewed ( 1 1 , 1 2). We do not attempt to review the voluminous literature on all aspects of mitosis and meiosis here, but concentrate on the fundamental problem of chromosome segregation: the nature of the mechanisms that ensure that sister chromatids (in mitosis and the second meiotic division) or homologous chromosomes (in meiosis I) segregate to opposite poles of the spindle. We emphasize studies on genetically tractable organisms, especi ally baker's yeast (Saccharomyces cerevisiae). We use the term kinetochore to describe the structural components that are assembled on the centro meric DNA, reserving the term centromere for the DNA itself. We use the term spindle pole body (SPB) to refer to the general concept of spindle pole organizing activity, although the morphological SPB is found only in fungi. In animal cells the centrosome is the microtubule organizing center. Mitosis and meiosis have been studied in many organisms using a wide variety of techniques. We have tried to integrate our discussion by con sidering the experimental evidence in the context of a specific model for chromosome segregation. The model rests on a synthesis of ideas that have all been put forward at least once by other workers. Six central concepts underlie the model: 1. Normal chromosome segregation is dependent on the integrity of microtubules (reviewed in 4 and 5). 2. Separate mechanisms exist that can generate force on chromosomes either towards or away from the spindle pole to which they are attached (reviewed in 4). 3. The stable attachment of a chromosome pair to the spindle requires the generation of forces that pull the two kinetochores of that chromosome pair towards opposite poles of the spindle, creating tension on a flexible linkage between the two chromosomal units to be segregated (13, 14). 4. The replication of DNA molecules can generate linkage of the daughter DNA molecules in a chromosome by intertwining (catenation) of the DNA duplexes (17). Catenation can be destroyed during anaphase by the activation of type II DNA topoisomerases, which catalyze the passage of one DNA duplex through another (reviewed in 1 8).

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

291

5. In most organisms normal segregation of homologs at meiosis I is dependent on genetic recombination (3, 1 5). 6. The difference between the behavior of sister chromatids in meiosis I from that in mitosis and meiosis II is the result of differences in the organization of the chromosomes rather than that of the spindle (16). A MODEL FOR CHROMOSOME BEHAVIOR DURING MITOSIS AND MEIOSIS
Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

The following model is highly speculative but has the advantages of pro viding a unified interpretation of a wide variety of data and of making testable predictions. The Mitotic Cell Cycle
INTERPHASE

Our model for the mitotic cell cycle is shown in Figure 1. At the beginning of the cell cycle each chromosome has a single kinetochore. We follow Sundin & Varshavsky in proposing that the recognition that a given pair of kinetochores are sisters is a result of the flexible linkage between the m produced by the intercatenation of sister chromatids (19). Intercatenation arises when duplex DNA is replicated in the absence of topoisomerase activity. Under these conditions each double-helical turn of the parental duplex is converted into one intertwining of the daughter molecules (Figure 2). Sundin & Varshavsky proposed that linkage arises at the points where replication forks meet as a result of the steric exclusion of DNA topoisomerase by the replication complexes (17). If intercatenation is stable from S to mitosis then each chromatid should be catenated to its sister at each of the many points where replication forks met. Alternatively, most intercatenation could be resolved during G2. In this case the catenation that we propose links sister chromatids together at metaphase could be generated by the late replication of the centromeric DNA (19). Tschumper & Carbon proposed that the yeast centromere acts as a block to the passage of replication forks (20). Release of this block at prophase would allow replication of the centromeric DNA with retention of sister chromatid linkage as a local catenation of the DNA at the centromere. Our proposal for the mechanics of chromosome movement during mitosis is based on a recent review by Pickett-Heaps et al (4). Microtubules in the two halves of the spindle have opposite polarities, and all those within one half have the same polarity (21 , 22). Because of the observed bipolarity of the spindle there are no fundamental difficulties in inventing schemes that will supply the vectorial chromosome movement required at anaphase.
METAPHASE

292

MURRAY & SZOSTAK

Pickett-Heaps et al (4) suggested that the kinetochores can interact with microtubules in two orientations. In the polar (P) orientation the kineto chore is pulled towards the spindle pole from which the microtubule originates. We suggest that attachment in the P orientation acts as a molecular ratchet: As long as the P force is opposed by a force acting in the opposite direction attachment is stable. If the kinetochores of two sister chromatids are attached in P orientation. to microtubules from opposite poles of the spindle, they will move towards the poles to which they are
Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

a
)

Figure 1

A model for mitotic chromosome segregation.

(a) DNA replication leads to two

sister chromatids attached to each other by catenation (each sister chromatid is represented b a single line).

(b)

Metaphase: stable attachment to the spindle is achieved when the

kinetochores are attached to opposite poles. Symbols:

(c) Anaphase: DNA topoisomerase activit)

decatenates the sister chromatids and allows them to move to opposite poles of the spindle.

triangle

kinetochore;

circle

spindle pole.

CHROMOSOME SEGREGATION

IN

MITOSIS AND MEIOSIS

293

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

Figure 2

Generation of catenation. When two replication forks meet the last 10-20 turns of

the DNA helix are converted into intertwinings of the daughter DNA duplexes.

attached until the linkage between them is stretched so that the P force on one kinetochore is opposed by that acting in the opposite direction on its sister. Attachment of a kinetochore to a microtubule in the opposite polarity yields the antipolar (AP) orientation. In this orientation the kinetochore generates force that moves it away from the pole from which the microtubule originates. Because all attachments at metaphase are in the P orientation, attachment in the AP orientation must be unstable. We propose that during metaphase the separation of sister chromatids is prevented by their catenation. We suggest that a feedback mechanism exists that activates DNA topoisomerase II only after all the kjnetochores are attached to microtubules in the P orientation. The activation of topoisomerase marks the transition from metaphase to anaphase, as it is the destruction of the catenation between sister
ANAPHASE

294

MURRAY & SZOSTAK

chromatids that allows them to separate and move polewards under the influence of the P force. Meiosis We argue that the meiotic cell cycle represents a mitotic cell cycle modified by the addition of a second duplication of the SPB without an intervening S phase, allowing two rounds of chromosome segregation to occur in a single cell cycle. Specifically, we propose that kinetochore duplication and centromere replication is delayed until the completion of meiosis 1. During meiosis II the centromere DNA is replicated, leaving the kinetochores attached to each other by intercatenation, exactly as they are in mitotic metaphase. We propose that the proper disjunction of homologs in meiosis I also requires intercatenation between segregating chromosomes, and that this linkage arises as the result of recombination. If the sister chromatids that make up a homolog are catenated at many points as a result of the collision of replication forks in meiotic S phase, then any reciprocal cross-over (but not gene conversion events) between nonsister chromatids will generate linkage of the homologs (Figure 3). After recombination each recombinant chromatid will be catenated to its sister centromere proximal to the point of recombination, and to its nonsister distal to the site of cross-over. (We define the sister of a recombinant chromatid as being the chromatid with which it shares a kinetochore in meiosis 1.) If homologs are linked by intercatenation, this linkage must be destroyed by the activation of a type II topoisomerase at the metaphase-anaphase transition of meiosis I. The topoisomerase activation will also destroy the intercatenation of sister chromatids. We propose that the linkage is regenerated at meiosis II as the result of replication of the centromeric DNA after the completion of meiosis 1. The delayed replication of the kinetochore's structural elements and the centromeric DNA means that at meiosis I the homologs each have only one kinetochore and no separation of sister chromatids can occur (Figure 3). In this scheme kinetochore duplication commits the cell to a mitotic division. Two modifications are required to allow meiosis to evolve from mitosis: (a) introduction of a precocious SPB duplication that allows meiosis I to occur before the mitosis-like meiosis II; and (b) the suppression of kinetochore duplication until after the completion of meiosis I. If the regulation of SPB duplication and spindle assembly in meiosis I is different from that in meiosis II and mitosis we expect to find mutations that block the assembly of the meiosis I spindle, leading to meiotic divisions in which recombination is normal but meiosis I fails to occur. Such mutants are in fact known (83).

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

295

MEIOSIS I

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

MEIOSlsn
Figure 3
A model for chromosome segregation in meiosis. In meiosis I two homologous pairs of sister chromatids are shown. Each sister chromatid pair has a single kinetochore and the centromeric DNA has not been replicated. (a) Recombination leads to the catenation of sister and nonsister chromatids, providing a flexible linkage between the two pairs of sister chromatids. (b) Metaphase I : stable attachment to the spindle occurs when the kinetochores are attached to opposite poles of the spindle. (c) Anaphase I : DNA topoisomerase destroys the catenation, allowing the sister chromatid pairs to segregate to opposite poles of the spindle.

(d) Replication of the centromeric DNA generates a pair of catenated sister chromatids (only one of the two products of meiosis I is shown). (e) Metaphase II; (f) Anaphase II : DNA
topoisomerase decatenates the sister chromatids allowing them to segregate to opposite poles of the spindle. Symbols:

triangle

kinetochore; circle

spindle pole.

EXPERIMENTAL EVIDENCE The Mitotic Cell Cycle We suggest that structural com ponents of the kinetochore, but not the centromeric DNA, are duplicated during G 1, although the timing of kinetochore duplication is not essential
TIME OF KINETOCHORE DUPLICATION

296

MURRAY & SZOSTAK

to the integrity of the model. There is no direct assay available for kinetochore duplication. Immunofluorescence studies on mammalian cells using antikinetochore sera from patients suffering from autoimmune diseases show that the kinetochores appear as single structures in most interphase cells (25, 26). In a small fraction of G2 cells the kinetochores appear as pairs of closely apposed dots (26). There are approximately the same number of paired dots as there are chromosomes in the cell line, suggesting that the staining reveals pairs of sister kinetochores. In the fungus Saproglenia, kinetochores are visible throughout the cell cycle and are always associated with microtubules (27). Electron micro scopy reveals that the morphological duplication of the kinetochore occurs immediately before mitosis (27). However, it is possible that the structural components of the kinetochore actually duplicate much earlier in the cell cycle, but that the morphological duplication is not visible until the centromeric DNA replicates during G2. Because the kinetochores of S. cerevisiae are not visible, even in electron micrographs, we do not know if they remain associated with microtubules throughout metaphase, although nuclear microtubules are present in this period (reviewed in 1 1). Immunofluorescence of yeast cells with antitubulin antisera does not show extensive microtubule arrays in the interphase nucleus (28), suggesting that in interphase the microtubules may extend only a short distance from the SPB. There is no compelling reason to argue against the duplication of the structural components of the kinetochore in S or in early G2, although it is interesting to note that during mitosis of cells in which endoreduplication has occurred the sister chromatids that arise from the extra round of DNA replication fail to segregate from each other (29). This suggests that this pair of DNA duplexes shares a single kinetochore, and thus that kinetochore duplication does not occur during S phase. The yeast cell cycle mutant cdc31 may also be defective in kinetochore duplication. At the nonpermis sive temperature the duplication of the spindle pole body, which normally occurs in Gl, is blocked (30). However, the SPB increases in size, and the processes of DNA replication and the initial stages of spindle formation occur as normal (30). The cells arrest in nuclear division, and electron microscopy shows that a normal spindle forms that has a SPB only at one end (30). This raises obvious questions about the role of the SPB in spindle formation. When cdc31 cells are grown at a semipermissive temperature their ploidy increases (31). This suggests that under these conditions sister chromatids fail to separate, so that all the chromosomes go to a single pole. We suggest that the CDC31 gene product is required for kinetochore as well as SPB duplication, consistent with a simultaneous duplication of the SPB and kinetochores in G1. One unresolved question is whether it is the

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

CHROMOSOME SEGREGATION

IN

MITOSIS AND MEIOSIS

297

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

mother or daughter cell that receives the chromosomes when a cdc31 strain divides at the semipermissive temperature. Another yeast gene product that may be required for kinetochore duplication is that of the NDC1 gene. In cold-sensitive ndc1 mutants the ploidy of the cells increases when they are grown at semipermissive temperatures. At the nonpermissive temperature pairs of cells accumulate in which one of the progeny inherits all of the DNA, as judged by staining with the DNA-specific dye DAPI, although the spindle appears normal by antitubulin immunofluorescence. The cell that inherits the DNA can be either the mother or the daughter cell (J. Thomas & D. Botstein, personal communication). The behavior of this mutant is compatible with the possibility that lack of the functional NDC1 gene product blocks kineto chore duplication so that sister chromatids fail to separate. Kinetochores may remain attached to one of the SPBs throughout interphase. This would provide an attractive explanation for the cosegrega tion of chromatids whose oldest strand is of the same age, which has been observed in yeast (32), Neurospora (33), and some mammalian cells (34). If kinetochore duplication was conservative such that the older kinetochores remained attached both to the oldest DNA strands and to one of the two SPBs, then all the old kinetochores would segregate to the same spindle pole, carrying with them the chromatids that contain the oldest DNA strands. In yeast the mother and daughter cells were equally likely to inherit the old set of chromatids (32), which suggests that the SPB attached to the old kinetochores is equally likely to be inherited by either cell ; this is consistent with our explanation of the behavior of the ndc1 mutant. The time at which different chromosomal regions replicate has been investigated in plant and animal cells. In general the centro meres and telomeres are late replicating (reviewed in 35). However, since these regions are also the sites of satellite DNA sequences (36), which are often late replicating (36), it is not possible to tell if the functional centromeres and telomeres are themselves late replicating.
CENTROMERIC DNA REPLICATION

Sundin & Varshavsky (17) de monstrated that the replication of the animal virus SV40 produced catenated dimers. They suggested that sister chromatids could also become catenated as a result of DNA replication, and that this might help direct mitotic chromosome segregation (19). Topoisomerase II was shown to be required for the separation of bacterial chromosomes in vivo, and catenated chromosomes could be induced to separate in vitro by the action of topoisomerase (38). Would the same result be obtained with eucaryotic chromosomes isolated from metaphase cells?
DNA REPLICATION AND CATENATION

298

MURRAY & SZOSTAK

In the yeast top2 mutant, which carries a temperature-sensitive DNA topoisomerase II, cells arrest during nuclear division at the nonpermissive temperature (39). In addition, the 2 p.m circle becomes highly intercatenated (39). No intercatenation of the circle is seen at the permissive temperature (39), which suggests that the half-life of catenated forms is short, and therefore that decatenation occurs prior to anaphase. If sister chromatid intercatenation, generated during S phase, survives from the end of S through metaphase, the removal of catenation by topoisomerase activity must occur much more slowly for long linear DNA molecules than it does for short circular ones. By performing shift-up and shift-down experiments on strains carrying a temperature-sensitive top2 mutation C. Holm & D. Botstein have shown that the activity of the TOP2 gene product is required at mitosis (personal communication). These experiments do not reveal whether topoisomerase II is also active at other points in the cell cycle. In the absence 01 topoisomerase activity the cells arrest at mitosis, and cell death rapidly ensues. The death of these mitotic cells can be prevented by adding nocodazole (a drug that inhibits tubulin polymerization) before shifting cells to the nonpermissive temperature (c. Holm & D. Botstein, personal communication). This suggests that death is due to chromosome breakage or aneuploidy that occurs when cells attempt to enter anaphase with catenated sister chromatids. When diploid yeast cells are treated for prolonged periods with the microtubule depolymerizing drug, methyl benzimadazol carbamate (MBC), a high fraction of aneuploid cells are seen in cells rescued on drug free medium (40). One interpretation of this finding is that aneuploidy arises because the catenation between sister chromatids slowly decays in cells that are arrested in mitosis, allowing a defective anaphase separation process to occur. If this idea is correct, top2 mutants should show a much lower frequency of aneuploidy when they are arrested in mitosis by MBC at the nonpermissive than at the permissive temperature. At metaphase in diatoms the sister kinetochores lie close to their respective poles, separated from each other by about 5-10 p.m (41). Thus, the proximity of the kinetochores cannot be what allows them to be recognized as sisters. The hypothesis that sister chromatid intercatenation normally holds daughter DNA molecules together at metaphase provides exactly the sort of flexible, long range connection this observation requires.
PROMETAPHASE CHROMOSOME MOVEMENT Pickett-Heaps and his collabo rators have made extensive studies of prometaphase chromosome move ment in diatoms, which led to the concepts of P and AP force (reviewed in 4). From experiments using colchicine and drugs that prevent ATP generation

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

299

they drew the following conclusions:

1. Attachment to the kinetochores in the P orientation leads to the

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

generation of P force, and the movement of the kinetochores towards the pole to which the microtubule is attached. P movement does not require ATP and can occur in the absence of microtubules (42, 43). 2. Attachment of the kinetochores to microtubules in the AP orientation leads to the generation of AP force, and the movement of the kinetochore away from the spindle pole to which the microtubule is attached. AP movement requires ATP and intact microtubules (42, 43). 3. In prometaphase the initial movement of the kinetochores is always polewards. This suggests that if the kinetochores remain attached to microtubules during interphase, they do so in the P orientation (41). Attachment of both kinetochores in the P orientation to opposite poles of the spindle does not necessarily specify the position of the chromosome on the spindle. The chromosome will come to rest only when the forces applied from opposite directions are equal. Ostergren (44) proposed that if the force on a kinetochore were proportional to its distance from the spindle pole then chromosomes would lie precisely between the poles at metaphase. This idea was tested by examining the position of trivalents (formed in spermatocytes from grasshoppers which had been treated with X-rays) at metaphase of meiosis I (45). At equilibrium the two kinetochores that face one pole must collectively generate no more force than the single kinetochore that faces the other pole. The kinetochore-to-pole distance for the single kinetochore was twice that for the two kinetochores that faced the opposite pole, as predicted by Ostergren (44). It is difficult to see how force generated at either the kinetochore or the spindle pole could be affected by the distance between them. This suggests either that P force is generated along the length of the microtubules, or alternatively that it represents some elastic component to which the kinetochore is attached, with the micro tubule acting as the guide rather than the agent that communicates force to the kinetochore. The energy independence and the microtubule indepen dence of P movement make the latter possibility attractive. An example of an elastic protein is the contractile protein spasmin found in certain rotifers. The rotifers are carried on contractile stalks composed largely of spasmin, which changes from a highly extended to a globular conformation in response to an increase in the calcium concentration. This conformational change produces a rapid contraction of the stalk (46). Possibly a similar elastic protein generates the P force in the spindle.
MICROMANIPULATION OF METAPHASE CHROMOSOMES The ability to micro manipulate chromosomes has made it possible to directly test the effect of

300

MURRAY & SZOSTAK

stretching the interkinetochore linkage on chromosome orientation and segregation. So far these experiments have been possible only in meiotic cells. In an elegant series of experiments Nicklas and his colleagues micromanipulated the chromosomes of grasshopper primary spermato cytes in metaphase of meiosis I (13, 14,47). The formation of chiasmata links the kinetochores of a pair of homologs,with the chromosome arms forming the linkage. By detaching one of the kinetochores of a pair of homologs from the spindle and moving it into the opposite half of the spindle, both kinetochores were attached to the same spindle pole (unipolar orientation). After this operation both kinetochores moved towards the pole to which they were attached, until one kinetochore detached from the spindle. The detached kinetochore would reattach at random to fibers from either spindle pole. Reattachment that maintained the unipolar orientation that resulted from the original manipulation was unstable. Only when the kinetochore reattached to the pole from which it had originally been detached did the two kinetochores move towards opposite poles and the chromosome orientation become stable. Such reorientation events were no more likely to occur near the spindle poles than at the equator (14). The reorientation could be prevented if the needle used for micromanipu lation was inserted between the homologs, and used to apply force towards the spindle pole that was not attached to the kinetochores ( 1 5). The simplest interpretation of this result is that spindle attachment becomes stable only when the linkage between the kinetochores is stretched. The high frequency of reorientation seen in spermatocytes recovering from cold treatment could also be prevented by stressing the linkage between the kinetochores (47). In addition, univalents (either sex chromosomes in the heterogametic sex, or autosomes that have no chiasmata) have been observed to reorient continually during the first meiotic division (48). The transition from metaphase to anaphase failed to occur in cells in which one chromosome had been detached from the spindle (13). This suggests that the cell is able to monitor the attachment of kinetochores to the spindle. As one example, the kinetochores could have a protein kinase activity that phosphorylates and thereby inactivates topoisomerase II. This activity might be expressed only in mitosis and be inhibited when the kinetochore is attached to the spindle in the P orientation. When all the kinetochores become attached in the P orientation, the inhibition of topo isomerase activity would be abolished and decatenation of sister chroma tids would accompany the transition from metaphase to anaphase.
ANAPHASE

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

In our model, anaphase is initiated by the decatenation of the sister chromatids, which allows each chromatid to move polewards as a result of P force. Indirect arguments suggest that the magnitude of the poleward forces increase during anaphase. Dicentric chromosomes are

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

301

broken in anaphase if the two kinetochores are attached to opposite poles of the spindle (49). In contrast, ring chromosomes that appear to be catenated in metaphase are not broken, which suggests that the magnitude of the polewards forces is smaller than in anaphase. At anaphase these ring chromosomes appear to pass through each other without breakage, an observation that strongly suggests the action of DNA topoisomerases (50). The fate of dicentrics suggests that chromosomes linked by catenation would be broken if cells made the transition to anaphase without decatenating sister chromatids. The force pulling sister chromatids apart could increase as a result of an increase in the magnitude of the P forces transmitted by the kinetochore-to-pole microtubules, or the elongation of the spindle mediated by interactions between the pole-to-pole micro tubules, or both. Force on the kinetochores does not seem to be required for the initial stages of sister kinetochore separation at anaphase. Sister chromatid separation occurs in cells treated with colchicine (51), in acentric fragments (51), and in monopolar spindles (52). These observations strongly suggest that some linkage between sister chromatids is destroyed as cells progress through mitosis. Our model postulates that the separation of sister chro matids is due to the activation of type II topoisomerase at the begin ning of anaphase. What mediates the transition between metaphase and anaphase, and the activation of topoisomerase? One attractive candidate is destruction of maturation promoting factor (MPF), an activity that was first identified in Xenopus by the ability of cytoplasm from hormonally activated oocytes to induce non-hormone-treated oocytes to pass through meiosis I and yield mature eggs (reviewed in 53). Since then MPF activity has been found in a wide variety of mitotic and meiotic cells. When injected into cells of early Xenopus embryos, MPF causes chromosome condensation, nuclear en velope breakdown, and spindle formation (54). Cells remain in metaphase until the level of MPF decays (55). Either MPF or some component required for its activation is destroyed at the end of mitosis, since new protein synthesis is required for the generation of MPF activity in the next cell cycle (56). The oscillation in MPF levels during the cell cycle probably represents a component of the cell cycle clock that is activated by ferti lization in Xenopus eggs (57). One way in which MPF could prolong metaphase is by acting either directly or indirectly to inhibit topoisomerase II activity. Perhaps attach ment of all the kinetochores to the spindle in the P orientation activates a system that destroys MPF activity.
ARTIFICIAL CHROMOSOMES

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

The availability of cloned replicators (58), telomeres (59, 60), and centromeres (61, 62) in yeast has opened up a new

302

MURRAY & SZOSTAK

approach to chromosome structure and behavior: the construction in vitro of molecules that contain all the elements known to be required for normal chromosome function. These artificial chromosomes are then introduced into yeast cells to examine their segregation in mitosis and meiosis (63, 64). The first linear artificial chromosomes were 1 0-15 kilobases (kb) in length (63, 64). Surprisingly, they were present in many copies per cell, and were much less mitotically stable than circular centromeric plasmids, which were present at one copy per cell and were lost about once in every hundred mitoses (61, 62, 64). The behavior of these linear plasmids was consistent with random segregation at mitosis, rather than the ordered segregation of sister molecules characteristic of circular centromeric plasmids or natural chromosomes. The defect in segregation is not due to mutations in the centromere (64), interference between the centromeric and telomeric chromatin structures (64), or the failure of the centromeres to attach to the mitotic spindle. The last point was demonstrated by constructing linear plasmids that contained two centromeres. Like dicentric circular plasmids (65), these linear molecules are structurally unstable, and they yield rearrangements that delete one ofthe two centromeres (Murray & Szostak, unpublished). Such deletions are the result of mechanical breakage that occurs after the two kinetochores of a dicentric molecule have attached to opposite spindle poles (49). Since one of the centromeres was only 2.5 kb from the end of the linear dicentric plasmid, it seems unlikely that there is a minimum telomere-centromere separation that must be exceeded before spindle attachment can occur. The short linear centromeric plasmids are less than 5% of the length of the smallest yeast chromosome [chromosome I is 300 kb (66)]. Larger artificial chromosomes (55 kb) show evidence of normal sister chromatid segregation: their copy number is one, and their mitotic loss frequency is about 10-2 (64). A further increase in length to 1 04 kb increases the stability by a factor of five, but such molecules are still at least one hundred-fold less stable than natural chromosomes (Murray & Szostak, unpublished). The 55 kb artificial chromosomes pair and segregate from each other normally in meiosis, although again the fidelity of segregation is substantially less than that of natural chromosomes (64).
DERIVATIVES OF NATURAL CHROMOSOMES The ability to create directed alterations in the structure of the yeast genome (67, 68) has made it possible to test the effect of various structural changes on chromosome behavior. The first type of alteration is created by integrating a linear telomere bearing plasmid into a yeast chromosome (Figure 4a) (Murray & Szostak, unpublished). Plasmid integration is directed by cleavage with a restriction enzyme that cuts in a region of the plasmid that is homologous to the

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

A) BREAKAGE
Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

B) SHORTENING

(. (
0

Figure 4

Manipulation of chromosome structure. (a) A chromosome can be broken and a new telomere generated by the directed introduction of a

0 ) ! ( 1 !
)
Resolution

<::

Integration

C>-

(')

Transformation

is

I
z

) (

AcentriC lost

CI!I

Recombination

CI!I

)
Dark bars represent homologous sequences on the plasmid and chromosome; arrows

!
w o w

plasmid carrying an inverted repeat of telomeric sequences. distant sites on the chromosome.

telomeric DNA sequences. (b) A chromosome arm can be shortened by transformation with a linear DNA fragment whose ends are homologous to

Open bars represent fragment sequences that are not homologous to the chromosome.

304

MURRAY & SZOSTAK

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

desired site of integration. The chromosome is broken into two fragments at the site of integration. The fragment that carries the centromere is mitotically stable and is retained, while the acentric fragment is rapidly lost. The experiments are conducted in diploid strains so that the loss of a chromosome fragment is not lethal. To test the effect of length on chromosome behavior we introduced telomeres at various points on yeast chromosome III. Centromeric chromosome fragments of 42 and 70 kb are lost mitotically at frequencies of 1 0-2 and 1 0-3 respectively, in comparison to chromosome 111 (530 kb) (66}, which is lost at a frequency of 1 0-5. Similar results have been obtained by Tye & Surosky (personal communication). For the smaller fragment, pedigree analysis has shown that 70% of the loss events occur by nondisjunction [the sister of the cell that lacks the chromosome fragment carries two copies (Murray & Szostak, unpublished)]. This is in marked contrast to the behavior of circular centromeric plasmids, in which only 25% of the events are due to nondisjunction (69, 70). The remaining events have only one copy in the sister cell that inherited the plasmid, which suggests that the plasmid failed to replicate in the cell cycle preceding loss, or that one copy of the plasmid was destroyed after replication. IncreaSing the size of circular centromeric plasmids from 5 to 1 00 kb causes a tenfold decrease in mitotic loss frequency from 1 0-2 to 10-3 (69). Unlike linear centromeric plasmids, the copy number of the circular centromeric molecules is one, irrespective of their size. A circular version of chromosome III is lost at a frequency of 10-3 (69, 71), and many of the events are due to sister chromatid exchange that creates a dicentric dimeric chromosome. Since sister chromatid exchange does not affect the structure of linear chromosomes, this may have been one of the selective forces favoring the evolution of linear chromosomes as genome size increased during evolution. A number of results suggest that the length of DNA molecules is one of the main determinants of mitotic stability: 1 . The mitotic loss frequencies of chromosome fragments and artificial chromosomes of similar sizes are within a factor of 5 of each other, despite the fact that most of the DNA on the artificial chromosomes is procaryotic (Murray & Szostak, unpublished). 2. Fragments of chromosome III of decreasing size show gradually decreasing stability, which demonstrates that there is no single element (apart from the centromere) whose presence is required for a high level of stability (R. Surosky & B.-K. Tye, personal communication; Murray & Szostak, unpublished). 3. The yeast mutant chll increases the spontaneous frequency of

CHROMOSOME SEGREGAnON

IN

MITOSIS AND MEIOSIS

305

chromosome loss (72). In this mutant the frequency of chromosome loss decreases with increasing chromosome size (72). 4. In yeast zygotes where nuclear fusion is prevented by the karl mutation, chromosomes are transferred between the two nuclei. The frequency of chromosome transfer decreases with increasing chromosome size (73). The second type of chromosome manipulation is the creation of large internal deletions, by transforming with a DNA fragment one of whose ends is homologous to a site on the chromosome, and the other end of which is homologous to some distant chromosomal site (Figure 4b) (R. Surosky & B.-K. Tye, personal communication; Murray & Szostak, unpUblished); A recombination event at both ends of the fragment replaces all the chromosomal information that lies between the two sites of homology with
the sequences of the fragment. This technique has been used to produce an

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

almost full length version of chromosome III whose centromere is only 3.5 kb from the nearest telomere. The telocentric chromosome is ten times less stable than the normal metacentric chromosome III (Murray & Szostak, unpublished). The telocentric chromosome is still one thousand times more stable than the 10-15 kb linear centromeric plasmids ; this demonstrates that the primary cause of the plasmids' instability is not the closeness of their centromeres to their telomeres. The results of studies on both artificial chromosomes and derivatives of natural yeast chromosomes suggest that the fundamental determinant of mitotic stability is the length of linear DNA molecules. This is consistent with the idea that catenation is an essential prerequisite for proper chromosome segregation. If the products of DNA replication are catenated circular molecules they are topologically linked to each other. In contrast, short linear molecules are not linked: The intertwinings can be resolved simply by twisting one molecule about the other, which explains why they, but not circular centromeric plasmids, segregate randomly at mitosis. To account for the ordered segregation of long artificial chromosomes we suggest that they can become linked to each other during replication. Such linkage requires the existence of topological barriers that constrain the catenation generated by replication. These barriers could be either attachment of the chromosome to the nuclear matrix, or the formation of chromosomal loops by DNA-binding proteins, which would noncovalently cross-link the DNA. Such domains of supercoiling have been seen in eucaryotic chromosomes, and it is intriguing that they are about 50 kb long (74), the size of the smallest artificial chromosome that shows ordered segregation at mitosis. The length of DNA molecules may also affect their ability to become

306

MURRAY & SZOSTAK

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

catenated due to the collision of replication forks. On a circular molecule a single origin of replication suffices to give catenation because the two forks arising from this origin will meet on the opposite side of the circle. For a linear molecule the forks will meet only if the plasmid has at least two replication origins, and both origins fire before either has been repli cated by the fork emanating from the other origin. The chance of fulfilling the second criterion increases as the distance between the origins in creases. In particular, linear molecules with a single replication origin should be incapable of generating intercatenation, and should show a very high frequen<:y of nondisjunction. Such molecules have not yet been constructed.
Meiosis

All the available evidence suggests that homolog disjunction is the result of some form of chromosome pairing (reviewed in 3, 9, 12). In most organisms pairing proceeds through the formation of homologous pairs of chromosomes joined' by the syn aptonemal complex (SC). Pairing is initiated in zygotene when the SC begins to form. This process has been studied in yeast (75), insects (12), mammals (12), and plants (76) by examining nuclei reconstructed from serial section electron micrographs. It appears that SC formation is initiated at the telomeres, which are in contact with the nuclear envelope, and then proceeds towards the centromeres. However, the presence of telomeres is not essential for SC formation and recombination since diploids containing two circular copies of chromosome III show normal levels of recombination for markers on chromosome III (71 ). By the beginning of pachytene the SC covers the full length of each bivalent. Two lines of evidence suggest that SC formation is an essential prerequisite for recombination in yeast: (a) In the cdc7 mutant, cells arrested at the nonpermissive temperature fail to form SC or commit to recombination, while both processes occur after shifting to the permissive temperature (77) ; and (b) arrest of wild-type yeast cells in pachytene by exposure to 36C leads to marked increases in recombination (78). SC formation is not sufficient for recombination, since female silkmoths (Bombyx) form apparently normal SC but fail to undergo recombination (12). The genetic analysis of meiotic recombination and the different models invoked to explain it have recently been reviewed elsewhere (79). In pachytene, small electron-dense structures are found associated with the SC, and these have been named recombination nodules (80) because (a) their number and distribution is roughly equal to the number and distribution of recombination events (75, 80); and (b) nodules are absent in organisms like female Bombyx that lack recombination, even if they have morphologically normal SC (1 2).
PAIRING AND DISJUNCTION

CHROMOSOME SEGREGAnON IN MITOSIS AND MEIOSIS

307

Although pairing appears to precede recombination, and recombination is dispensable in certain organisms [male Drosophila (reviewed in 3), female Bombyx], a number of lines of evidence suggest that recombination is normally involved in ensuring homolog segregation. Recombination results in the formation of chiasmata that act as points of linkage between homologs. As a result, the kinetochores of a pair of homologs are physically linked but can be widely separated unless the chiasmata lie close to the centromere. In both Drosophila (81) and yeast (E. Lambie & S. Roeder, personal communication) meiotic recombination near the centromere is suppressed. In yeast, moving the centromere of chromosome III from its normal location near the LEU2 gene to the HIS4 gene increases the frequency of recombination near LEU2 and decreases it near HIS4; this demonstrates that the centromere itself is responsible for suppressing recombination (E. Lambie & S. Roeder, personal communication). One interpretation of this suppression of recombination is that it ensures that the kinetochores are not too tightly linked, and makes it easy for the two kinetochores of a pair of homologs to capture microtubules from opposite poles of the meiotic spindle. To achieve stable bipolar orientation an unattached kinetochore must find a microtubule with a polarity opposite the one to which its sister is attached. A flexible interkinetochore linkage will allow the unattached kinetochore to scan a larger nuclear volume than if it were rigidly attached to its sister. While it seems clear that chiasmata link homologous chromosomes that have recombined, the nature,of the linkage is unknown. In maize there is an exact correlation between the number of chiasmata and the number of reciprocal exchanges (reviewed in 82). This suggests that recombination events that are resolved as cross-overs give rise to chiasmata, while those that are resolved as gene conversions do not. This observation implies that the sister chromatids of homologs are attached to each other along their length. Thus, a cross-over will leave the recombinant chromatids catenated to their sister's centromere proximal to the point of exchange, and to their nonsister's centromere distally (Figure 3). Gene conversion, which merely replaces the information on one chromatid with a short stretch from its nonsister, would not lead to this linkage. As the homologs separate the chiasmata move toward the telomeres: The linkage of nonsister chromatids by catenation explains this phenomenon. Topoisomerase-mediated removal of intertwining at the chiasmata allows the homologs to move apart under the influence of the polewards force exerted at the kinetochores. However, removal of intertwinings at sites other than the chiasmata fails to lead to separation because the homologs remain linked at the chiasmata. Thus the chiasmata move progressively towards the ends of the chromosomes as if topoisomerase acted preferen tially at the chiasmata.

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

308

MURRAY & SZOSTAK

The mutation spo13 blocks the first meiotic division, and diploid spores are produced (83). These spores show normal levels of recombination, which demonstrates that this mutant has the expected phenotype for a failure of the "extra" SPB duplication invoked in the model for the meiotic cell cycle described previously. The duplication of SPBs in the mitotic cell cycle is dependent on the CDC31 gene product. To explain the behavior of the spo13 mutant we postulate that:
YEAST MEIOTIC MUTANTS
1. The CDC31 gene product can induce both SPB and kinetochore duplication, while the SPO13 gene product can induce SPB but not kinetochore duplication. 2. The normal mitotic cell cycle incorporates a feedback mechanism that prevents the SPBs from being duplicated twice between rounds of chromo some segregation. 3. The activity of the SPO13 and CDC31 gene products are regulated in a mutually exclusive fashion: CDC31 is active in mitotically grown cells and in meiosis II, while SP013 is active only during meiosis I. 4. During meiosis I the continued activity of the SPO13 gene product is dependent on maintenance of the conditions required to induce sporulation.

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

The expression of the SPO13 gene product is known to be induced by sporulation medium (R. E. Esposito, personal communication). According to our scenario, cells starting the cell cycle in sporulation media will have high levels of SP013 gene product, and will duplicate their SPBs, but not their kinetochores, which will lead to meiosis I. The completion of meiosis I will allow the CDC31 function to become active, kinetochore and SPB duplication will occur, and meiosis II (which is mechanistically identical to mitosis) will follow. If cells are returned to growth conditions before meiosis I occurs the CDC31 function will become active, the kinetochores will be duplicated, and meiosis I will be replaced by a mitotic division. The diploid products of this division will have suffered meiotic levels of recombination. Such cells are observed when sporulating cultures are returned to vegetative growth (24). Thus our model explains why cells cannot enter the meiotic cell cycle after SPB duplication, and how they can escape from the meiotic cell cycle before meiosis I. The effects of the cdc31 and ndc1 mutants on meiosis have been examined. When strains mutant for either gene are put through meiosis at the nonpermissive temperature two diploid spores are recovered, and genetic analysis indicates that homologous chromosomes have separated at meiosis I, but that meiosis II and the segregation of sister chromatids has failed to occur (30, J. Thomas & D. Botstein, personal communication). This behavior strongly suggests that meiosis II is functionally equivalent to

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

309

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

mitosis, and is consistent with the idea that these functions are required for kinetochore duplication, and that kinetochore duplication in meiosis does not occur until after the completion of meiosis I. SP01 3 (wild-type) cells that contain the recombination deficient muta tions spol l , rad50, or rad52 produce inviable spores (reviewed in 9, 15). Electron microscopic studies show that the rad50 mutant lacks syn aptonemal complex, while spol l possesses it (B. Byers & R. E. Esposito, personal communication), which demonstrates that the formation of SC is not sufficient to ensure homolog disjunction in yeast. In the case of spoI l a small number of viable spores can be recovered, and these have not undergone meiotic recombination (84). The spol l , spo13 double mutant produces normal levels of viable diploid spores that have not undergone recombination (84). The presence of spoI l increases spore viability in spol 3 diploids, which suggests that meiotic recombination events can interfere with chromosome segregation in the second (mitosis-like) division (84). Perhaps the linkage of homologs interferes with their correct orientation on the metaphase plate, or with the segregation of sister chromatids at anaphase. The spol 3, rad50 double mutant also produces viable but unrecombined spores (1 5). This suggests that both RAD50 and SPOI l gene products are required to initiate meitoic recombination, and that the failure to do so leads to the failure of homolog disjunction at meiosis I, and thus to inviable spores. In Drosophila a number of meiotic mutants that produce inviable and aneuploid gametes have been shown to be deficient for meiotic recombination (reviewed in 3). In female Drosophila those homologs (or unpaired chromosomes) that have not recombined enter a second pairing system, the distributive pairing pool, in which chromo somes disjoin from each other on the basis of size (reviewed in 85). The dis tributive pairing mechanism is apparently sufficient to deal with about four unpaired chromosomes. Higher numbers lead to its breakdown, and very high frequencies of nondisjunction (3). In male Drosophila, which lack homologous pairing, recombination, and distributive pairing, the homo logs appear to pair via specific pairing sites in the heterochromatin (reviewed in 3). The Drosophila meiotic mutant pal leads to chromosome loss in meiosis, and chromosome loss in the early divisions of the zygote (86). pal affects meiosis only in males, and the chromosomes that are lost from the zygote nucleus and its progeny are always the paternal chromosomes (86). In one stock the X chromosome was much more resistant to pal-induced loss. In crosses this resistance to loss maps to the centromeric heterochromatin of the X (86). We suggest that a wild-type product of the pal gene acts to
DROSOPIIILA MEIOTIC MUTANTS

310

MURRAY & SZOSTAK

suppress kinetochore assembly before meiosis I in males. The mutant alleles allow kinetochore duplication to occur before meiosis I ; therefore, the newly assembled kinetochores are defective because some substance or condition required for kinetochore assembly is absent at this stage of meiosis. In meiosis II this would lead to chromosomes with one normal and one defective kinetochore, and a high frequency of nondisjunction. Simultaneous loss of both chromosomes X and 4 is more frequent than expected from the product of the frequencies of exceptional segregation for either one alone (86). The behavior of the two chromosomes is correlated. In the case of early mitotic loss both chromosomes are almost always lost from the same pole in division (86). This behavior is reminiscent of the yeast ndc1 mutant in that the chromosomes that nondisjoin all arrive at the same pole in mitosis and meiosis. However, in the ndc1 mutant either all or none of the chromosomes in a cell nondisjoin and segregate to a single pole, while in the pal mutant some of the chromosomes disjoin and others segregate normally. The correlated behavior of chromosomes could be explained if the defective kinetochores assembled in the meiotic cell cycle always face the same pole in subsequent divisions. One way in which such coorientation can be achieved is to attach all the kinetochores of the same age to the same pole. If our interpretation is correct, the behavior of these mutants argues that kinetochores of the same age are all attached to one pole in meiosis II as well as in mitosis. Goldstein (87) has shown that in wild-type males at prometaphase of meiosis I the kinetochore of each homolog is a single hemispherical structure. By late metaphase the kinetochore has metamorphosed into a double structure that consists of two discs lying side by side (87). While these two discs could be functionally independent they have the same orientation, and thus will attach to microtubules of the same polarity, so there will be no tendency of sister chromatids to separate. We predict that electron microscopy of meiosis in pal males would disclose the occurrence of kinetochore duplication before meiosis I. Are there meiosis-specific genes that prevent sister kinetochore separa tion before metaphase of meiosis II? Defects in such genes would be ex pected to give rise to precocious separation of sister chromatids, which would be scored as nondisj unction at either meiosis I or meiosis II, depending on when separation occurred. Two mutants with this property have been described in Drosophila, mei-S322 (88) and ord (89). They are unique among meiotic mutants in that they act in both males and females, and both have been cytologically observed to cause the precocious separation of sister chromatids in male meiosis (90). The fraction of nondisjunction events that occurs at meiosis II is 95% and 30% for mei-S322 and ord, respectively. For mei-S322 the frequency of gametes that carry

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

CHROMOSOME SEGREGATION

IN

MITOSIS AND MEIOSIS

311

n o copies of a given chromosome is much greater than that of gametes that carry two copies of the same chromosome, which suggests that some chromosomes are not included in the nuclei that form after the completion of meiosis.

MEIOSIS I VERSUS MEIOSIS II

Sister chromatids separate from each other in

mitosis and meiosis II, but not in meiosis I. Does this reflect differences in the structure of the meiosis I and II spindles, or is it a reflection of the organization of the chromosomes themselves? One test between these alternatives is provided by an elegant experiment of Nicklas (16). He induced primary and secondary spermatocytes to fuse with each other to produce a single cell that had both meiosis I and meiosis II spindles. He then transferred either a bivalent from the meiosis I to the meiosis II spindle, or a pair of sister chromatids from the meiosis II to the meiosis I spindle. In both cases the transferred entities behaved as they would have done on the spindle they originally belonged to : The bivalent separated into two homologs, while the sister chromatids from meiosis II separated from each other (16). This result is entirely consistent with our argument that kinetochore duplication does not occur until the interval between meiosis I and II. Homologs transferred from the meiosis I spindle would have a single kinetochore, while those transferred from the meiosis II to meiosis I spindle would have duplicated kinetochores, allowing the sister chromatids to disjoin from each other.

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

EVOLUTION OF MEIOSIS
Our model provides a unified view ofthe mitotic and meiotic cell cycles, and suggests how meiosis evolved from mitosis. In the primitive forms of meiosis, disjunction was probably ensured by synaptonemal complex formation without recombination. Although in mitosis the chromosomes were flexibly attached to each other by intercatenation, the rigid association of homo logs in meiosis mediated by the SC probably led to a high frequency of nondisjunction. The introduction of recombination to create a flexible linkage between homologs would have increased the fidelity of chromo some segregation. Thus the initial selection for the evolution of meiotic recombination may have been for a decrease in nondisjunction, rather than an increase in the potential for genetic diversity in the succeeding generation.

ACKNOWLEDGMENT
Because of space limitations we were unable to quote every reference to a given topic. To those whose names are not mentioned we offer our

312

MURRAY & SZOSTAK

apologies. We are extremely grateful to all those who provided information before publication. We thank D. Dawson, P. Szauter, and B. Alberts for their helpful comments on the manuscript, and T. Claus for photography.
Literature Cited
1. Hartwell, L. H., Dutcher, S. K., Wood, 1. S., Garvik, B. 1982. The fidelity of mitotic chromosome reproduction in S. cere visiae. Rec. Adv. Yeast Mol. BioI. 1 : 2838 2. Sora, S., Luchini, G., Magni, G. E. 1982. Meiotic diploid progeny and meiotic non-disjunction in Saccharomyces cere visiae. Genetics 101: 1 7-33 3. Baker, B. S., Hall, J. C. 1976. Meiotic mutants : genic control of meiotic recom bination and chromosome segregation. In : Genetics and Biology o Drosophila, f Vol. I, ed. E. Novitski, M. Ashburner, pp. 351-429. New York : Academic 4. Pickett-Heaps, J. D., Tippit, D. H., Porter, K. R. 1982. Rethinking Mitosis. Cell 29: 729-44 5. Inoue, S. 1 981 . Cell division and the mitotic spindle. J. Cell Bioi. 91 : 1 3 1 s-47s 6. Cabral, F. 1 984. Genetic dissection of the assembly of micro tubules and their role in mitosis. Cell. Muscle Moti!. 5: 3 1 3-40 7. Hepler, P. K., Wolniak, S. M. 1984. , Membranes in the mitotic apparatus : their structure and function. Int. Rev. Cytol. 90: 169-238 8. Rieder, C. L. 1982. The formation, struc ture and composition of the mammalian kinetochore and kinetochore fiber. Int. Rev. Cytol. 79 : 1-58 9. Esposito, R. E., Klapholz, S. 1981. Meiosis and ascospore development. In
.

1 5.

16.

17.

18. 19.

20.

21.

22.

Molecular Biology o the Yeast Saccha f romyces, Lif Cycle and Inheritance, e

10.

1 1. 12.

13.

14.

ed. J. Strathern, E. W. Jones, J. R. Broach, pp. 2 1 1-88. Cold Spring Harbor, New York : Cold Spring Harbor Lab. Baker, B. S., Carpenter, A. T. c., Esposito, M. S., Esposito, R. E., Sandler, L. 1976. The genetic control of meiosis. Ann. Rev. Genet. 10:53-134 Byers, B. 198 1 . Cytology of the yeast life cycle. See Ref. 9, pp. 59-96 Rasmussen, S. W., Holm, P. B. 1980. Mechanics ofmeiosis. Heriditas 93: 187216 Nicklas, R. B . 1967. Chromosome micro manipulation U. Induced reorientation and the experimental control of segrega tion in meiosis. Chromosoma 2 1 : 1 7-57 Nicklas, R. B., Koch, C. A. 1 969. Chromosome micromanipulation III.

23. 24.

25.

26.

27. 28.

Spindle fiber tension and the reorien tation of maloriented chromosomes. J. Cell BioI. 43: 40-50 Malone, R. E., Easton-Esposito, R. 1 98 1 . Recombinationless meiosis i n Saccha romyces cerevisiae. Mol. Cell. Bioi. 1 : 891-901 Nicklas, R. B. 1977. Chromosome distri bution : experiments on cell hybrids and in vitro. Philos. Trans. R. Soc. London B 277: 267-76 Sundin, 0., Varshavsky, A. 1980. Terminal stages of SV40 DNA repli cation proceed via multiply intertwined catenated dimers. Cell 21 : 103-14 Gellert, M. 198 1 . DNA topoisomerases. Ann. Rev. Biochem. 50: 879-91 0 Sundin, 0 . , Varshavsky, A . 1981. Arrest of segregation leads to accumulation of highly intercatenated dimers : dissection of the final stages of SV40 DNA repli cation. Cell 25: 659-69 Tschumper, G., Carbon, 1. 1983. Copy number control by a yeast centromere. Gene 23: 221-32 Telzer, B. R., Haimo, L. T. 1 98 1 . Decoration o f spindle microtubules with dynein : evidence for uniform polarity. J. Cell BioI. 89: 373-78 Euteneur, U., McIntosh, J. R. 1 98 1 . Structural polarity o f kinetochore micro tubules in PtKI cells. J. Cell Bioi. 89: 338-45 Deleted in proof Sherman, F., Roman, H. 1963. Evidence for two types of allelic recombination in yeast. Genetics 48: 255-61 Moroi, Y., Hartman, A. L., Nakane, P. K., Tan, E. M. 1981. Distribution of kinetochore (centromere) antigen in mammalian cell nuclei. J. Cell Bioi. 90 : 254--5 9 Brenner, S., Pepper, D., Berns, M. W., Tan, E., Brinkley, B. R. 1982. Kine tochore structure duplication and distri bution in mammalian cells : Analysis by autoantibodies from scleroderma patients. J. Cell BioI. 91: 95-102 Heath, I. B. 1980. Behavior of the kineto chores in the fungus Saproglenia ferax. J. Cell Bioi. 84 : 5 3 1 Kilmartin, J . V., Adams, A. E. M. 1984. Structural rearrangements of actin and

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

CHROMOSOME SEGREGATION IN MITOSIS AND MEIOSIS

tubulin during the cell cycle of the yeast Saccharomyces. J. Cell BioI. 98 : 922-33 29. Brenner, S., Branch, A., Meredith, S., Berns, M. W. 1977. The absence of centrioles from spindle poles of rat kan garoo (PtK2) cells undergoing meiotic like reductional division in vitro. J. Cell BioI. 72 : 368-79 30. Byers, B. 198 1 . Multiple roles of the spindle pole bodies in the life cycle of Saccharomyces cerevisiae. In Molecular

313

Studies on kinetochore function in mi tosis. II. The effects of metabolic inhibi tors on mitosis in the diatom Hantzchia amphioxus. Eur. J. Cell Bioi. 28 : 83-91 43. Pickett-Heaps, J. D., Spurck, T. P. 1982. Studies on kinetochore function in mi tosis. I. The effects of cytochalasin and colchicine on mitosis in the diatom 28 : 77-82 44. Ostergren, G. 1950. Considerations on some elementary features of mitosis. Heriditas 36 : 1-19 45. Hays, T. S., Wise, D., Salmon, E. D. 1982. Traction force on a kinetochore fiber acts as a linear function of kinetochore fiber length. J. Cell BioI. 93 : 374-82 46. Amos, W. B., Routledge, L. L., Yew, F. F. 1975. Calcium binding proteins in a vorticellid contractile organelle. J. Cell " Sci. 19 : 203-1 3 47. Henderson, S . A., Nicklas, R . B., Koch, C. A. 1970. Temperature induced orien tation instability during meiosis : an ex perimental analysis. J. Cell Sci. 6 : 323-50 48. Nicklas, R. B. 196 1 . Recurrent pole to pole movements of the sex chromosome of Melanoplus diff erentialis in spermato cytes. Chromosoma 1 2 : 97-1 1 5 49. McClintock, B . 1938. The production of homozygous deficient tissues with mutant characteristics by means of the aberrant mitotic behavior of ring shaped chromosomes. Genetics 23 : 3 1 5-76 50. Bajer, A., Ostergren, G. 1963. Obser vations on transverse movement within the phragmoblast. Hereditas 50 : 1 79-95 5 1 . Bajer, A., Mole-Bajer, J. 1963. Cine analysis of some aspects of mitosis in endosperm. In Cinematography in Cell Biology, ed. G. Rose, pp. 357-409. New York : Academic 52. Bajer, A. S. 1982. Functional anatomy of monopolar spindle and evidence for oscillatory movement in mitosis. J. Cell Bioi. 93 : 33-48 53. Matsui, Y., Clarke, H. 1. 1979. Oocyte maturation. Int. Rev. Cytol. 57 : 1 85-282 54. Miake-Lye, R., Newport, J. W., Kirschner, M. 1983. Maturation pro moting factor induces nuclear envelope breakdown in cycloheximide arrested embryos of Xenopus laevis. J. Cell Bioi. 97 : 8 1-91 55. Newport, J. W., Kirschner, M. 1 984. Regulation of the cell cycle during early Xenopus development. Cell 37 : 73 1-42 56. Gerhart, J., Wu, M., Kirschner, M. 1 984. Cell cycle dynamics of an M-phase speci fic cytoplasmic factor in Xenopus laevis oocytes and eggs. J. Cell Bioi. 98 : 124755

Hantzchia amphioxus. Eur. J. Cell Bioi.

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

31.

32.

33.

34.

35.

36.

37. 38.

39.

40.

1 6, ed. D. von Wettstein et aI, pp. 1 19-34. Copenhagen : Munksgaard Schild, D., Ananthaswarmy, H. N., Mortimer, R. K. 1 98 1 . An endomitotic effect of a cell cycle mutation of Saccha romyces cerevisiae. Genetics 97 : 55 1-62 Williamson, D. H., Fennell, D. J. 198 1 . Non-random assortment o f sister chro matids in yeast mitosis. See Ref. 30, pp. 89-102 Rosenburger, R. F., Kessel, M. 1968. Non-random sister chromatid segrega tion and nuclear migration in hyphae of Aspergillus nidulans. J. Bacterial. 96 : 1 208- 1 3 Lark, K. G . , Consigli, R. A., Minocha, H . C . 1966. Segregation o f sister chromatids in mammalian cells. Science 1 54 : 1 202-4 Back, F. 1976. The variable condition of euchromatin and heterochromatin. Int. Rev. Cytol. 45 : 25-64 Appels, R., Peacock, W. 1. 1978. The arrangement and evolution of highly repeated (satellite) DNA sequences with special reference to Drosophila. Int. Rev. Cytol. Suppl. 8 : 70-127 Deleted in proof Steck, F. R., Drilca, K. 1984. Bacterial chromosome segregation : evidence for DNA gyrase involvement in decatena tion. Cell 36 : 1081-88 DiNardo, S., Voelkel, K. A., Sternaglanz, R. L. 1 984. DNA topoisomerase mutant of S. cerevisiae : topoisomerase II is re quired for segregation of daughter mole cules at the termination of DNA repli cation. Proc. Natl. Acad. Sci. USA 8 1 : 2616-20 Wood, 1. S., Hartwell, L. H. 1982. A dependent pathway of gene functions leading to chromosome segregation in

Genetics in Yeast. Alf red Benzon Symp.

94 : 71 8-26 41. Tippet, D. H., Pickett-Heaps, J. D., Leslie, R. L. 1980. Cell division in two large pennate diatoms Hantzchia and Nitzchia III. A new proposal for kineto chore function during prometaphase. J. Cell BioI. 86 : 402-1 6 42. Pickett-Heaps, J . D., Spurck, T . P. 1982.

Saccharomyces cerevisiae. J. Cell BioI.

3 14

MURRAY

&

SZOSTAK
Haber, J. E. 1978. Characterization of a mutation in yeast causing non-random chromosome loss during mitosis. Genetics 88 : 650-71 73. Dutcher, S. K. 198 1 . Internuclear trans f of genetic information in kari/KARI er heterokaryons in Saccharomyces cere visiae. Mol. Cell. Bioi. 1 : 245-53 74. B'anyajati, e., Worcel, A. 1976. Isolation, characterization and structure of the folded interphase chromosome of Dro sophila melanogaster. Cell 9 : 393-407 75. Byers, B., Goetsch, L. 1975. Electron microscopic observations on the karyotype of diploid and tetraploid 76. Gillies, e. B. 1984. The synaptonemal complex in higher plants. Crit. Rev. Plant. Sci. 2 : 81-1 1 6 77. Schild, D., Byers, B . 1978. Meiotic effects of DNA-defective cell divisio cycle mutants of Saccharomyces cerevisiae. Chromosoma 70 : 109-30 78. Davidow, L. S., Byers, B. 1984. Enhanced gene conversion and post-meiotic segre gation in pachytene arrested Saccharo myces cerevisiae. Genetics 106 : 165--8 3 79. Orr-Weaver, T. L., Szostak, J. W. 1985. Fungal recombination. Microbiol. Rev. 49 : 33-58 SO. Carpenter, A.T.e. 1975. Electron micro scopy of meiosis in Drosophila melano gaster females. II. The recombination nodule-a recombination associated structure of pachytene. Proc. Natl. Acad. Sci. USA 72 : 3186--90 81. Carpenter, A. T. e. , Baker, B. S. 1982. The control ofthe distribution of meiotic exchange in Drosophila melanogaster. Genetics 101 : 8 1-90 82. John, B., Lewis, K. R. 1976. The Meiotic Mechanism (Oxford Biology Readers), ed. J. J. Head. Oxford : Oxford Univ. Press 83. Klapholz, S., Esposito, R. E. 1 980. Recombination and chromosome segre ation during the single division meiosis III spol2-l and spo13-l diploids. Genetics 96 : 589-61 1 84. Wagstaff, J. E., Klapholz, S., Esposito, R. E. 1982. Meiosis in haploid yeast. Proc. Natl. Acad. Sci. USA 72 : 2986--90 85. Grell, R. F. 1976. Distributive pairing. In Genetics and Molecular Biology o f Drosophila, Vol. 1, pp. 435-85, ed. E. Novitski, M. Ashburner. New York : Academic 86. Baker, B. S. 1975. Paternal loss (pal) : a meiotic mutant in Drosophila melanogas-

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.

57. Hara, K., Tydeman, P., Kirschner, M. 1 980. A cytoplasmic clock with the same period as the division cycle in Xenopus eggs. Proc. N atl. Acad. Sci. USA 77 : 46266 58. Struhl, K., Stinchcomb, D. T., Scherer, S., Davis, R. W. 1979. High frequency trans formation of yeast : Autonomous repli cation of hybrid DNA molecules. Proc. Natl. Acad. Sci. USA 76 : 1035--39 59. Szostak, J. W., Blackburn, E. H. Cloning yeast telomeres on linear plasmid vec tors. Cell 29 : 245-55 60. Szostak, J. W. 1982. Structural require ments for telomere resolution. Cold Spring Harbor Symp. Quant. BioI. 47 : 1 1 8 7-94 61. Clarke, L., Carbon, J. 1980. Isolation of a yeast centromere and construction of functional small circular chromosomes. Nature 287 : 504-9 62. Stinchcomb, D. T., Mann, e., Davis, R. W. 1982. Centromeric DNA from 158 : 157-79 63. Dani, G. M., Zakian, V. A. 1983. Mitotic and meiotic stability oflinear plasmids in yeast. Proc. Natl. Acad. Sci. USA 80 : 3406-- 1 0 64. Murray, A. W., Szostak, J. W. 1983. Construction of artificial chromosomes in yeast. Nature 305 : 1 89-93 65. Mann, C., Davis, R. W. 1983. Instability of dicentric plasmids in yeast. Proc. N atl. 66. Schwartz, D. e., Cantor, e. R. 1984. Separation of chromosome sized DNAs by pulsed field gradient gel electropho resis. Cell 3 7 : 67-95 67. Orr-Weaver, T. L., Szostak, J. W., Rothstein, R. J. 1983. Genetic appli cations of yeast transformation with linear and gapped plasmids. Methods Enzymol. 1 0 1 : 228-45 6S. Rothstein, R. J. 1983. One step gene disruption in yeast. Methods Enzymol. 101 : 202-1 1 69. Hieter, P., Mann, C., Snyder, M., Davis, R. W. 1985. Mitotic stability of yeast chromosomes : a colony assay that measures non-disjunction and chromo some loss. Cell 40 : 381-92 70. Koshland, D., Kent, J. C., Hartwell, L. H. 1985. Genetic analysis of the mitotic transmission of mini-chromosomes. Cell 40 : 393-403 71. Haber, J. E., Thornburn, P. C., Rogers, D. 1984. Meiotic and mitotic behaviour of dicentric chromosomes in Saccha romyces cerevisiae. Genetics 106 : 185-205

72. Liras, P., McCusker, J., Masciola, S.,

Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 72 : 5056--60

Saccharomyces cerevisiae. J. Mol. Bioi.

Acad. Sci. USA 80 : 228-32

CHROMOSOME SEGREGAnON
ter causing loss o f paternal chromo somes. Genetics 80 : 267-96
87. Goldstein, L. S. B. 198 1 . Kinetochore structure and its role in chromosome orientation during the first meiotic divi sion in male Drosophila melanogaster. Cell 25 : 591-602 88. Davis, B. K. 1 97 1 . Genetic analysis of a meiotic mutant resulting in precocious sister centromere separation in Droso

IN

MITOSIS AND MEIOSIS

315

1 1 3 : 251-72 89. Mason, J. M. 1976. Orientation disrup tor (ord) : a recombination defective and disjunction defective meiotic mutant in Drosophila melanogaster. Genetics 84 : 545-72 90. Goldstein, L. S. B. 1 980. Mechanisms of chromosome segregation revealed by two meiotic mutants of Drosophila mela nogaster. Chromosoma 78 : 79- 1 1 1

phila melanogaster. Mol. Gen. Genet.

Annu. Rev. Cell. Biol. 1985.1:289-315. Downloaded from arjournals.annualreviews.org by HARVARD UNIVERSITY on 07/16/09. For personal use only.